AI in Australian contact centres: what the data tells us
Is AI really transforming contact centres? We explore the 2026 Best Practice Report to reveal true adoption rates, process barriers, and data challenges.
It feels like every vendor at every conference right now is pitching the exact same AI story, promising massive transformations and the 'contact centre of the future.'
But let's look at what is really going on. The 2026 Contact Centre Best Practice Report surveyed over 600 people across the Australian industry. Kaizn presents a chapter at the SMAART Best Practice Roadshow each year, and this was our fifth. We covered the topic AI for self service, the first year that this has been surveyed to the Australian contact centre industry.
Here’s a quick summary of what the data showed.
Most contact centres are in the 'watching' phase
44% of contact centres describe their current AI position as "only analysing calls and data" - the single most common response, by a significant margin. Only 4% are scaling end-to-end automation across multiple workflows.
Only 2% said they're not interested or unable to implement AI. The rest are watching, reviewing vendors, running pilots in corners of their operation — but not committing. In almost every conversation we have with contact centre leaders, the reason is the same: they don't want to make the wrong call in a market that moves this fast, and the cost of a bad decision feels higher than the cost of waiting.
The barriers are people and process problems
When asked what are the biggest barriers preventing you from automating more of your contact centre workflows, budget limitations top the list at 54%.
What else is limiting automation:
- Technical implementation skills — 55%
- Process design — 44%
- Governance and compliance expertise — 43%
- Change management — 42%
More software doesn't fix a process design gap. What this tells us is that the organisations struggling to scale aren't failing because they picked the wrong platform - they're failing because they went into implementation without the process foundations in place. A half-built pilot with no clear owner and no plan for change management is where most AI projects actually die, and it's got nothing to do with the technology.
Data readiness is a huge problem no one is yet confronting
3% of contact centres are completely confident in their data foundations.
AI automation, AI agents, AI self-service -all of it runs on data. Unreliable data produces unreliable outcomes, and unreliable outcomes kill internal confidence to scale.
Adrienne Merlo from Customer Driven put it plainly in the report: "You can't automate a broken process or fuel a genius tool with garbage data."
Organisations treating data readiness as a parallel workstream will find their AI ambitions harder to achieve and more expensive. Fix this first before anything.
Half the industry can't answer the most basic question about their self-service
We asked contact centres how many of their customers they're actually self-serving. 41% had no idea.
Think about that for a second. In an industry where almost everything is measured — queue times, handle times, abandonment rates — nearly half of us can't answer a basic question about one of our biggest technology investments. And if you can't answer it internally, you certainly can't answer it when your CFO looks across the table and asks what's actually coming through the channel before signing off on the next round of investment. You don't have a benchmark. You don't have a baseline. You've got an AI tool and a hunch.
Getting clearer on that number isn't just good housekeeping. It's the thing that makes every future conversation with the board easier.
Where AI is delivering results
Post-call automation has the most traction:
- Automatic call summarisation — 35% using it, strong momentum
- Real-time agent assist — the during-call technology with the highest combined adoption and experimentation rate
- Speech and interaction analytics — 22% using, 29% experimenting
These are the use cases producing early wins. High-visibility, lower-complexity, and the results show up immediately for frontline agents.
For AI self-service specifically:
- 53% of contact centres are using some form of self-service AI
- Two thirds have built it to handle basic information retrieval
- 13% achieve a completely seamless AI-to-human handover
- 77% ask customers to repeat information when escalating to a human agent
The foundations are there. The scope of what AI self-service is being asked to do remains limited.
Voice AI: strategy over curiosity
71% of contact centres expect their use of Voice AI to increase over the next two years.
Right now, 80% have no Voice AI deployed. Of those who use it, 60% describe it as somewhat or highly effective, while 40% are undecided or too early to tell.
Voice AI is moving from a point of curiosity to a point of strategy. The organisations building capability now will be better placed when deployment becomes standard
The window is narrowing
The report is direct: the groundwork laid in 2026 will determine who is best placed to scale in the years ahead. For contact centres still in observation mode, catching up without incurring significant costs becomes harder as automation becomes standard.
Daniel Harding is Founder and Director of Kaizn, an independent CX and AI advisory helping contact centres across ANZ make better technology decisions.
Want to talk through where your organisation sits with AI automation, AI agents, or AI self-service? Book a conversation here.
.png)
